Do quantum computers make money now?
HSBC partnered with IBM recently to demonstrate that quantum computers could increase bond trading success by up to 34%. But was this legit?
On the 25th of September, HSBC announced that quantum computers could be used within the sector of algorithmic trading. Specifically algorithmic bond trading. This was a hybrid approach, meaning that part of the process was controlled by a normal classical computer - whilst the other part was controlled by a quantum one. By doing this, the team were able to improve their predictions of how likely a trade would get ‘filled’. This means that the buyer and seller actually agree on a price and the transaction goes through.
Algorithmic trading in itself is a huge industry. By 2030, the sector is estimated to reach a value of $42.99 billion, growing at an annual rate of 12.9%. Virtually all forms of trading are becoming algorithmic. In the corporate bond market, one can use computer models to automatically set bids and make trades at incredibly fast rates. By automating this process, where every microsecond wasted could mean a loss, a company can maximise their profits and maintain competitive margins. One of the main hurdles to overcome in algorithmic trading is the design of algorithms that include every possible factor. This complexity is probably why quantum computers were first considered to be applied to bond trading. And according to IBM/HSBC, they managed to actually improve the current standard.
How did they do it?
In the process of algorithmic bond trading, the algorithms calculate the optimal prices to buy and sell assets. Essentially this is like two computers trading between each other. During this exchange, one of the parties’ algorithms will calculate how likely a certain bond price will result in a purchase. In order to train and evaluate the quantum computing method, the team used real trading data on multiple IBM quantum computers. They could then predict the probability of successful trades within the European bond market.
Vice President of IBM - Jay Gambetta - said:
“This exciting exploration shows what becomes possible when deep domain expertise is integrated with cutting-edge algorithm research, and the strengths of classical approaches are combined with the rich computational space offered by quantum computers.”
But it’s an odd result really…
It’s odd because if you even check the official paper, they state something which would send alarm bells ringing for any quantum computing physicist.
“These empirical results suggest that the inherent noise in current quantum hardware contributes to this effect and motivates further studies.”
This doesn’t really make much sense. Noise in itself is an error caused by the environment, known for disrupting quantum data and reducing the effect of quantum operations. Noise is random and never benefits quantum computation. Yet this paper claims that noise has actually improved their results.
To me this is akin to saying: “We let our quantum state randomly corrupt, and it worked??”.
I will admit… in certain circumstances, noise can improve quantum computation. But as a general rule, noise destroys quantum states. I wrote about this in an earlier article about decoherence:
So what can this all mean?
One of my favourite bloggers, Scott Aaronson, called this whole thing a ‘Qombie’. A quantum-zombie that roams the earth claiming quantum advantage when the advantage is all but dead. And he’s got a point. There aren’t many examples of overwhelming quantum advantage where it has actually been implemented in real life. Grover’s Algorithm, Shor’s Algorithm and Deutsch-Jozsa’s Algorithm have all either not been implemented due to hardware limitations or don’t have enough relevant practical uses.
Although there are a lot of roaming Qombies, nevertheless even bad press can be good. In this particular case, it is questionable as to whether this method of using quantum computers actually yields better results. Even so, large press-releases like this can drive funding and innovation towards the sector of quantum computing. This can allow for more capable quantum computers to be manufactured, which in turn should generate promising results.
